How our game model's projected score has graded against the market and the final result: straight-up winners, against the spread, and totals over/under.
Propaholics NFL model track record: every game graded three ways against the closing line, straight-up winner, against the spread, and total over/under, plus how far the projected score landed from the final. The model is an independent projection built only from play-by-play, never the betting lines, and the record is shown honestly, season by season.
16 games. Average projection error: 12.33 pts on the margin, 12.91 pts on the total. Pushes are excluded from win %.
| Date | Matchup | Final | Proj | Winner | Spread | Total |
|---|---|---|---|---|---|---|
| Sep 14, 26 | DEN @ KC | 10-31 | 21.7-22.6 | KC✓ | DEN +2.5✗ | Over 42.5✗ |
| Sep 13, 26 | ARI @ LAC | 26-14 | 20-24.4 | LAC✗ | ARI +8.5✓ | Under 47.5✓ |
| Sep 13, 26 | ATL @ PIT | 13-20 | 20.9-23.7 | PIT✓ | ATL +6.5✗ | Over 40.5✗ |
| Sep 13, 26 | BAL @ IND | 41-23 | 25.6-24.4 | BAL✓ | IND +3✗ | Over 48.5✓ |
| Sep 13, 26 | BUF @ HOU | 36-31 | 23.7-22.3 | BUF✓ | HOU +1.5✗ | Over 44.5✓ |
| Sep 13, 26 | CHI @ CAR | 59-37 | 24.9-21.7 | CHI✓ | CHI -3✓ | Under 47.5✗ |
| Sep 13, 26 | CLE @ JAX | 10-34 | 14.4-22.6 | JAX✓ | CLE +8.5✗ | Under 40.5✗ |
| Sep 13, 26 | DAL @ NYG | 20-28 | 24.4-25.7 | NYG✓ | NYG +3✓ | Over 47.5✓ |
| Sep 13, 26 | GB @ MIN | 22-39 | 22.5-20.9 | GB✗ | GB +1.5✗ | Under 46.5✗ |
| Sep 13, 26 | MIA @ LV | 13-27 | 21.2-18.9 | MIA✗ | MIA +3✗ | Under 40.5✓ |
| Sep 13, 26 | NO @ DET | 30-31 | 18.6-26.9 | DET✓ | DET -7✗ | Under 49.5✗ |
| Sep 13, 26 | NYJ @ TEN | 23-10 | 19.4-22.1 | TEN✗ | TEN -1.5✗ | Over 39.5✗ |
| Sep 13, 26 | TB @ CIN | 27-33 | 24.6-25.6 | CIN✓ | TB +3.5✗ | Under 50.5✗ |
| Sep 13, 26 | WAS @ PHI | 22-24 | 20.6-28 | PHI✓ | PHI -6✗ | Over 43.5✓ |
| Sep 10, 26 | SF @ LA | 27-7 | 23.5-29.3 | LA✗ | LA -3.5✗ | Over 47.5✗ |
| Sep 9, 26 | NE @ SEA | 10-13 | 20.1-23.6 | SEA✓ | SEA -3P | Under 44.5✓ |
The cards at the top show the model's straight-up winner record, its against-the-spread record, and its over/under record versus the closing line, plus the average error between the projected and actual score. It covers every game the model has called live this season, and grows each week.
The closing spread and total from the market, and the actual final score. The winner pick is graded straight up. The spread pick is graded against the closing number, and the total pick against the closing total, with pushes excluded from the win percentage.
It is an independent second opinion, not a market-beater. It picks straight-up winners at a solid clip but does not consistently beat the closing spread, which is the hardest benchmark there is. The record here is shown honestly so you can judge it for yourself.
The model is deterministic from play-by-play using pre-game power ratings, so every completed game's pick can be recomputed and graded. That means the record is a faithful account of what the model projected entering each game, refreshed as new games finish.